Re: attribute name conflicts

paul c wrote:

> I believe Codd originally envisaged "tables" that had what he called
> domain names to identify "columns" but that in his second paper, he
> introduced what we know of as attribute names because he wanted, for
> example, to allow for relations between things of the same type, such as
> ones used for bills of materials.
>
> That's fine as far as single relations go, but it's always led me to a
> question about a whole database. Does it ever make sense within a given
> application (as opposed to within a given db) to have two different
> attribute names that identify different types/domains?
>
> (obviously those attribute names would occur in the def'ns of distinct
> relations.)
>
> p

Did you perhaps mistype and mean: "Does it ever make sense within a given application (as opposed to within a given db) to have two different *attributes with identical names* that identify different types/domains?"

If so, I would say: Yes, absolutely. While I would expect the much more common case is for similarly named attributes to have the same type, I can imagine all sorts of cases where one might use the same name for different types of things.

Capacity might mean a volume in one case, a weight in another case and stored energy in a third case. Received on Wed Jun 27 2007 - 18:32:31 CEST